© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 45595
    Prev
    Next
    INDEX
    Explanations

    phrases indicating transitions or movements, particularly upward or in a positive direction

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    373
    +0.19
    1.0%
    314
    +0.10
    0.5%
    80
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    750
    +0.19
    0.04
    373
    +0.10
    -0.02
    340
    +0.09
    0.03
    Negative Logits
     Citation
    -0.71
    cases
    -0.54
     boycot
    -0.51
    cause
    -0.48
    isphere
    -0.48
    KS
    -0.48
    terday
    -0.48
     2024
    -0.47
     inclusion
    -0.47
    anian
    -0.47
    POSITIVE LOGITS
    ruff
    0.61
    ensed
    0.60
    senal
    0.58
     shenan
    0.58
    gettable
    0.53
    razy
    0.52
     proverbial
    0.51
     snipp
    0.51
     uncanny
    0.50
     tricks
    0.50
    Activations Density 0.215%

    No Known Activations